Mastering Visual Continuous Control: Improved Data-Augmented Reinforcement Learning

TL;DR

DrQ-v2 is a new model-free reinforcement learning algorithm that significantly improves visual continuous control, achieving state-of-the-art results on complex tasks with efficient training from pixel data.

Contribution

It introduces DrQ-v2, an improved, simple, and computationally efficient RL algorithm that advances visual control capabilities, including humanoid locomotion from pixels.

Findings

Achieves state-of-the-art results on DeepMind Control Suite

Solves complex humanoid tasks directly from pixels

Requires only 8 hours of training on a single GPU

Abstract

We present DrQ-v2, a model-free reinforcement learning (RL) algorithm for visual continuous control. DrQ-v2 builds on DrQ, an off-policy actor-critic approach that uses data augmentation to learn directly from pixels. We introduce several improvements that yield state-of-the-art results on the DeepMind Control Suite. Notably, DrQ-v2 is able to solve complex humanoid locomotion tasks directly from pixel observations, previously unattained by model-free RL. DrQ-v2 is conceptually simple, easy to implement, and provides significantly better computational footprint compared to prior work, with the majority of tasks taking just 8 hours to train on a single GPU. Finally, we publicly release DrQ-v2's implementation to provide RL practitioners with a strong and computationally efficient baseline.